Scoring profile with weights and a function - azure-cognitive-search

Question

I'm using Azure Search with a scoring profile. I need text fields along with quantity sold to be a part of the scoring profile. I can configure the following profile, but the quantity sold doesn't seem to be factored in to the search score when I query the index. I'm thinking because quantity sold isn't a string, its an int. Therefore, I can't make the field searchable? I'm using the new featuresMode parameter in the query, the quantity sold field doesn't even appear in the scoring breakdown
"scoringProfiles": [
{
"name": "Product Name",
"functions": [
{
"fieldName": "QuantitySold",
"freshness": null,
"interpolation": "linear",
"magnitude": {
"boostingRangeStart": 0,
"boostingRangeEnd": 100000,
"constantBoostBeyondRange": true
},
"distance": null,
"tag": null,
"type": "magnitude",
"boost": 6
}
],
"functionAggregation": "sum",
"text": {
"weights": {
"ProductName": 4,
"ProductSet": 3,
"ProductDesc": 2
}
}
}
],

Answer 1

What type of boost you should use depends on the datatype. An int like QuantitySold should use type magnitude for boosting. A date would use type freshness and so on.
I recreated a minimal example with the simplest possible index with only two properties: Id and Title.
CREATE INDEX
{
"#odata.context": "https://{{SEARCH_SVC}}.{{DNS_SUFFIX}}/$metadata#indexes/$entity",
"#odata.etag": "\"0x8D8761DCBBCCD00\"",
"name": "{{INDEX}}",
"defaultScoringProfile": null,
"fields": [
{
"name": "Id",
"type": "Edm.String",
"facetable": false,
"filterable": true,
"key": true,
"retrievable": true,
"searchable": true,
"sortable": true,
"analyzer": null,
"searchAnalyzer": null,
"synonymMaps": [],
"fields": []
},
{
"name": "Title",
"type": "Edm.String",
"facetable": false,
"filterable": true,
"key": false,
"retrievable": true,
"searchable": true,
"sortable": true,
"analyzer": null,
"searchAnalyzer": null,
"synonymMaps": [],
"fields": []
},
{
"name": "QuantitySold",
"type": "Edm.Int32",
"facetable": true,
"filterable": true,
"key": false,
"retrievable": true,
"searchable": false,
"sortable": true,
"analyzer": null,
"searchAnalyzer": null,
"synonymMaps": [],
"fields": []
}
], "scoringProfiles": [
{
"name": "relevance",
"text": {
"weights": {
"Title": 1.5
}
}
},
{
"name": "sales",
"functions": [
{
"type": "magnitude",
"fieldName": "QuantitySold",
"boost": 100,
"interpolation": "linear",
"magnitude": {
"boostingRangeStart": 1,
"boostingRangeEnd": 100000,
"constantBoostBeyondRange": false
}
}
]
} ],"corsOptions": null, "suggesters": [], "analyzers": [], "tokenizers": ], "tokenFilters": [], "charFilters": [], "encryptionKey": null}
UPLOAD MINIMAL
I then submit two products. One called Apple iPhone with a low sales quantity. And Apple Juice with a high sales quantity.
{
"value": [
{
"#search.action": "mergeOrUpload",
"Id": "2",
"Title": "Apple Juice",
"QuantitySold": 10000
},
{
"#search.action": "mergeOrUpload",
"Id": "1",
"Title": "Apple iPhone",
"QuantitySold": 35
}
]
}
QUERY
Without using any scoring profile, I query for apple. As expected, the two items are equally relevant responses to my query. They both match one of two tokens. Both get score 0.25811607
https://{{SEARCH_SVC}}.{{DNS_SUFFIX}}/indexes/{{INDEX}}/docs?search=apple&$count=true&searchMode=all&queryType=full&api-version={{API-VERSION}}&featuresMode=enabled
{
"#odata.count": 2,
"value": [
{
"#search.score": 0.25811607,
"#search.features": {
"Title": {
"uniqueTokenMatches": 1.0,
"similarityScore": 0.25811607,
"termFrequency": 1.0
}
},
"Id": "2",
"Title": "Apple Juice",
"QuantitySold": 10000
},
{
"#search.score": 0.25811607,
"#search.features": {
"Title": {
"uniqueTokenMatches": 1.0,
"similarityScore": 0.25811607,
"termFrequency": 1.0
}
},
"Id": "1",
"Title": "Apple iPhone",
"QuantitySold": 35
}
]
}
QUERY WITH BOOST ON QUANTITY SOLD
I then repeat the query for apple, but this time I boost items with a high QuantitySold by selecting my scoring profile called sales. This boosts our Apple Juice item to the top with a score of 2.8123235. The Apple iPhone item has also received a boost, but much smaller with only a score of 0.26680434.
https://{{SEARCH_SVC}}.{{DNS_SUFFIX}}/indexes/{{INDEX}}/docs?search=apple&$count=true&searchMode=all&queryType=full&api-version={{API-VERSION}}&featuresMode=enabled&scoringProfile=sales
{
"#odata.count": 2,
"value": [
{
"#search.score": 2.813235,
"#search.features": {
"Title": {
"uniqueTokenMatches": 1.0,
"similarityScore": 0.25811607,
"termFrequency": 1.0
}
},
"Id": "2",
"Title": "Apple Juice",
"QuantitySold": 10000
},
{
"#search.score": 0.26680434,
"#search.features": {
"Title": {
"uniqueTokenMatches": 1.0,
"similarityScore": 0.25811607,
"termFrequency": 1.0
}
},
"Id": "1",
"Title": "Apple iPhone",
"QuantitySold": 35
}
]
}